What GPU do you need to run Nemotron 3.5 Super VL 120B A12B?
124.45B params 1M max context VisionTool useReasoning
The smallest GPU memory that runs Nemotron 3.5 Super VL 120B A12B well is 72 GB (Q4_K_M, e.g. RTX PRO 5000 Blackwell 72GB). 15 of 132 GPUs run it well. It's a mixture-of-experts model, so all weights must fit in memory but speed depends only on the active experts.
SWE-bench Verified 71.1
VRAM needed per quantization
| Quantization | Weights | 16K context | 32K context | Full 1M |
|---|---|---|---|---|
| Q4_K_M | 65.3 GB | 65.4 GB | 65.5 GB | 73.5 GB |
| Q8_0 | 119.7 GB | 119.8 GB | 119.9 GB | 127.8 GB |
Weights plus fp16 KV cache. Leave ~1 GB headroom for the inference engine.
Cheapest GPUs that can run Nemotron 3.5 Super VL 120B A12B
Lowest launch MSRP among GPUs that run its smallest quantization (Q4_K_M) well, with at least 16K context.
As an Amazon Associate, I earn from qualifying purchases.
GPUs that can run Nemotron 3.5 Super VL 120B A12B
Best quantization that fits on each GPU, with the context and speed it gets. For chips with several memory sizes, the smallest that works is shown.
NVIDIA
- RTX PRO 5000 Blackwell 72GB 72 GB Q4_K_M 838K context · ~184 tok/s
- RTX PRO 5500 Blackwell 84 GB Q4_K_M 1M context · ~192 tok/s
- RTX PRO 6000 Blackwell 96 GB Q4_K_M 1M context · ~245 tok/s
- DGX Spark (GB 10) 128 GB Q8_0 1M context · ~22 tok/s
- RTX Spark 128 GB Q8_0 1M context · ~24 tok/s
AMD
- Ryzen AI Max (Strix Halo) 96 GB Q4_K_M 1M context · ~35 tok/s
- Ryzen AI Max 400 (Gorgon Halo) 128 GB Q8_0 1M context · ~22 tok/s
Apple
- M2 Max 96 GB Q4_K_M 1M context · ~55 tok/s
- M3 Max 96 GB Q4_K_M 1M context · ~41 tok/s
- M3 Ultra 96 GB Q4_K_M 1M context · ~112 tok/s
- M5 Ultra 96 GB Q4_K_M 1M context · ~164 tok/s
- M1 Ultra 128 GB Q8_0 1M context · ~64 tok/s
- M2 Ultra 128 GB Q8_0 1M context · ~64 tok/s
- M4 Max 128 GB Q8_0 1M context · ~44 tok/s
- M5 Max 128 GB Q8_0 1M context · ~49 tok/s